international journal of hydrogen energy 40 (2015) 11740e11747

Available online at www.sciencedirect.com ScienceDirect

journal homepage: www.elsevier.com/locate/he

Selectivity and resistance to poisons of commercial hydrogen sensors

* V. Palmisano a, E. Weidner a, , L. Boon-Brett a, C. Bonato a, F. Harskamp a, P. Moretto a, M.B. Post b, R. Burgess b, C. Rivkin b, W.J. Buttner b a Energy Conversion and Storage Unit, European Commission e DG Joint Research Centre Institute for Energy and Transport, Westerduinweg 3, (P.O.Box 2), 1755 ZG, Petten, The Netherlands b Transportation and Hydrogen Systems Center, National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, CO, 80401-3305, USA article info abstract

Article history: The resistance of several models of catalytic, workfunction-based metal-oxide-semi-

Received 19 November 2014 conductor and electrochemical hydrogen sensors to chemical contaminants such as SO2,

Received in revised form H2S, NO2 and hexamethyldisiloxane (HMDS) has been investigated. These sensor platforms 23 February 2015 are among the most commonly used for the detection of hydrogen. The evaluation pro- Accepted 25 February 2015 tocols were based on the methods recommended in the ISO 26142:2010 standard. Perma-

Available online 20 March 2015 nent alteration of the sensor response to the target analyte (H2) following exposure to potential poisons at the concentrations specified in ISO 26142 was rarely observed. Keywords: Although a shift in the baseline response was often observed during exposure to the po- Hydrogen sensors tential poisons, only in a few cases did this shift persist after removal of the contaminants. Hydrogen safety Overall, the resistance of the sensors to poisoning was good. However, a change in

Cross-sensitivity sensitivity to hydrogen was observed in the electrochemical platform after exposure to NO2

Poisons and for a catalytic sensor during exposure to SO2. The siloxane resistance test prescribed in Inhibitors ISO 26142, based on exposure to 10 ppm HMDS, may possibly not properly reflect sensor Interferents robustness to siloxanes. Further evaluation of the resistance of sensors to other Si-based contaminants and other exposure profiles (e.g., concentration, exposure times) is needed. Copyright © 2015, The Authors. Published by Elsevier Ltd on behalf of Hydrogen Energy Publications, LLC. This is an open access article under the CC BY-NC-SA license (http:// creativecommons.org/licenses/by-nc-sa/4.0/).

and fuel cell technologies depends directly on their perceived Introduction safety. Therefore hydrogen safety measures have been developed to ensure the safe use of hydrogen, including Hydrogen sensors are an important enabling technology for guidelines for mitigation of fault and accidents. Hydrogen the safe implementation of the emerging hydrogen infra- safety sensors can indicate hydrogen concentrations before structure. Hydrogen sensors are deployed to increase safety in the lower flammability limit (LFL) of 4 vol% in air is reached. applications such as hydrogen production, storage, distribu- Sensors are used to trigger an alarm, which may be followed tion and use. The market acceptance of emerging hydrogen by additional measures such as closing off the hydrogen

* Corresponding author. E-mail address: [email protected] (E. Weidner). http://dx.doi.org/10.1016/j.ijhydene.2015.02.120 0360-3199/Copyright © 2015, The Authors. Published by Elsevier Ltd on behalf of Hydrogen Energy Publications, LLC. This is an open access article under the CC BY-NC-SA license (http://creativecommons.org/licenses/by-nc-sa/4.0/). international journal of hydrogen energy 40 (2015) 11740e11747 11741

supply, increasing ventilation or initiating system shutdowns. oxidation of hydrogen in EC sensors typically takes place on a This is a significant contribution to the safe use of hydrogen. Pt/C catalytic layer [8]. The Pd or Pt catalyst of CAT sensors is To facilitate the reliable and proper use of hydrogen sensors commonly coated onto the alumina bead containing the fila- sensor testing facilities were independently established by the ment. MOSFET hydrogen sensors use platinum, palladium or European Commission's Joint Research Centre - Institute for an alloy containing these metals as the catalytic gate material Energy and Transport (IET) [1] and by the US Department of deposited as a thin film on an insulating oxide layer. Energy (DOE) at the National Renewable Energy Laboratory This catalyst may be susceptible to contaminants, which (NREL) [2]. may influence the response of the sensor to hydrogen. The Many types of hydrogen safety sensors are commercially sensors tested, as listed in Table 1 were selected based on their available, based on different mechanisms to detect hydrogen. proven robust performance, high level of development, and These sensors generally provide for a reliable detection of widespread deployment. hydrogen under a broad range of ambient conditions, how- In this work, the performance of these sensors during ever, performance gaps have been identified regarding some exposure to potential poisons is evaluated. This is a continu- applications [3]. Hydrogen sensors are widely implemented in ation of previous work, which analyzed the effect of potential industrial applications [4], but the deployment of hydrogen interferents on these and other sensor types [9]. Detailed de- safety sensors in novel applications may lead to different and scriptions of the detection principles of the various hydrogen more challenging performance requirements. Sensor selec- detection platforms has been presented elsewhere, e.g. tivity and robustness against poisons are especially important Ref. [7]. The specific contaminants used in this study were in applications where numerous chemical species could be chosen because they are listed in ISO 26142 as species to present and the exposure of the hydrogen sensor to a different which the resistance to poisoning of hydrogen detection species, i.e. contaminants, can lead to false alarms. As an apparatus needs to be evaluated for certification (see Table 2) example, sensors deployed in hydrogen refueling stations are [5]. likely to be exposed to NOx and SOx from the internal com- Sensors based on catalyzed chemical reactions (such as bustion engines of conventional vehicles. The lubricants and CAT and EC) make use of noble metal catalysts (e.g. Pd, Pt) sealants used in warehouses may be harmful to sensors which may be susceptible to catalyst poisoning. A poisoning mounted close to indoor refueling points for materials effect on the catalyst may be due to blocking of an active site, handling vehicles. Contaminants may also temporarily or affect the adsorption of other species, or the chemical nature permanently alter the sensor's response to hydrogen, which of the catalyst through the formation of new compounds [10]. could have serious safety consequences as leaked hydrogen The interaction between a potential poison and the catalyst may go undetected. depends in part on the electronic configuration of the species Selectivity describes the ability of a sensor to respond to involved, which controls both the formation and orientation the target analyte without influence from the presence of of chemical bonds between a poison and the catalyst. Ele- contaminants. A gas sensor developed for a specific target ments of the nitrogen (N, P, As, Sb) and oxygen (O, S, Se, Te) analyte (e.g. hydrogen) should not respond to other species groups act as poisons on platinum group metal catalysts [10]. that may be incidentally present at the point of use. The The availability of electrons for bonding can also explain the sensitivity of the hydrogen sensor to other gases is referred to order of increasing poisoning activity for sulphur species, as as cross-sensitivity. When the presence of a chemical other H2S has a stronger effect that SO2 [11]. than the target gas induces a temporary change in the sensor The effect of catalyst poisons on the performance of gas response, it is termed an interferent, whereas chemicals sensors is well known and counter-measures have been which permanently affect a sensor response to the target developed by sensor manufacturers. Different design strate- analyte are termed poisons. A hydrogen sensor fails the gies have been employed by sensor manufacturers to mini- requirement of ISO 26142 when its response to hydrogen mize cross-sensitivity and improve sensor resistance to varies by more than 20% as a result of exposure to a poisons (e.g. Ref. [12] for metal-oxide conductometric sen- contaminant [5]. It is noted that a contaminant may be a sors). The sensor stability will depend on the properties of poison on one sensor platform but may be an interferent or the catalyzing material and on the presence of filters or even inert on another. In this paper we report the resistance to protective membranes such as molecular sieve coatings. The poisoning of catalytic, metal-oxide-semiconductor and elec- interferents would otherwise attach to and block the active trochemical hydrogen sensor platforms evaluated to pro- sites of the catalyst inhibiting the hydrogen oxidation reac- cedures and requirements laid out in ISO 26142. This work was tion, which corresponds to the basic detection principle of a initiated under the auspices of a JRC-NREL Memorandum of catalytic sensor. A physical barrier can protect the catalyst Agreement [6]. material by preventing poisoning species from reaching it. There are a great variety of commercially available For example, hydrogen-permeable films of polytetrafluoro- hydrogen sensing platforms. The most commonly deployed ethylene [13] or fluorinated ethylene propylene [14] deposited platforms are electrochemical (EC) and catalytic pellistor on or above the catalytic surfaces have been used to prevent (CAT) sensors [7]. A relatively new platform for hydrogen the diffusion of potential poisons and interferents to the sensing is the workfunction-based metal-oxide semi- catalyst. An outer zeolite layer has been proposed for a cat- conductor sensor (MOS). While these platforms have distinc- alytic sensor to trap larger molecules before reaching the gas tively different hydrogen detection mechanisms, they all sensing element [15], as well as active charcoal or other filter share a common feature: all use a catalyst material for the materials [16]. For electrochemical sensors, membranes or dissociation or combustion of hydrogen. The electrochemical diffusion barriers have been used to improve selectivity as 11742 international journal of hydrogen energy 40 (2015) 11740e11747

Table 1 e Overview of tested sensors. Sensor technology Acronym Reported measuring range Reported cross-sensitivity Catalytic CAT-1 0e1.4 vol% Response to combustible gases (Combustible gas sensor) CAT- 2 1e4 vol % e Electrochemical EC 0 4 vol % No response to NO2,H2S and SO2 e Workfunction based semiconductor-metal-oxide MOS 0 4.4% vol % No response to H2S, NO2

well as reduce the effect of poisons [8]. Such physical barriers concentration of all test gases was calculated from the ratio of are usually based on a size exclusion effect, but chemical the respective gas flow rates as controlled by mass flow barriers have also been proposed. The use of a secondary controllers. catalyst material, providing strong redox sites to react with The exposure profile used for the cross-sensitivity test is poisons, but not catalysing gas combustion is described in a illustrated in Fig. 1 and consists of the following stages: US patent [17]. In general sensor manufacturers treat the identity and nature of the catalyst as well as the protective (a) Operation in clean air, followed by exposure to 1 vol% H2 measures as proprietary. in air and subsequent recovery in clean air (initial response test based on ISO 26142). The sensor response to this initial exposure to hydrogen was scaled so as to Experiments and procedures indicate 1 vol% in the following plots. (b) Exposure to the contaminant at the concentration specified in Table 2 followed by simultaneous exposure The impact of exposure to SO2,H2S, NO2 and HMDS on the to 1 vol% H in air and subsequent recovery in the performance of several hydrogen sensor platforms is reported 2 contaminant gas mixture (poisoning test based on ISO in this work. Four representative commercial sensor 26142). platforms. (c) The exposure sequence in stage (a) was repeated to (4 sensor models total) were selected. Two representative highlight any short term influence on the sensor's models of the catalytic platform were tested, with CAT-1 response to hydrogen following exposure to the corresponding to a low-cost model and CAT-2 to a more contaminant (final response test based on ISO 26142). complex model of combustible gas sensor. Neither model was listed by the manufacturer as being poison resistant. Table 2 indicates the concentrations of potential poisons used in this study. The concentrations were selected close to Cross-sensitivity testing procedure the levels specified in the ISO 26142 standard. Hydrogen con- centration was set to 1 vol%, which is 25% of the LFL and a The impact of chemical species as potential poisons on common alarm set point. hydrogen sensors was evaluated in a dedicated test fixture, which was designed and built for testing the performance of hydrogen sensors exposed to contaminants. The tests are Cross-sensitivity/poisoning assessment performed by exposing the sensors to gas flows alternating between hydrogen (1 vol%) and the contaminants as well as a The effect of the contaminants on the sensor response was combination of hydrogen with the contaminant at controlled evaluated by comparing the sensor response to air (baseline concentrations. Environmental parameters in the test cham- response) and to 1 vol% hydrogen in air in the presence and ber were maintained at 25 C ± 2 C and 100 kPa ± 2 kPa. Dry absence of the contaminant. In order to quantify the cross- test gases were used in experiments so that the relative hu- sensitivity and the poisoning effect of the different sensors midity was typically less than 5%. The total gas flow rate in the tested, the cross-sensitivity factors, X0 and XH, and the test chambers was 1000 sccm. The desired test gas mixtures poisoning factor XP were defined, X0 was defined as the ratio of net sensor response to the contaminant and the net sensor were generated by dynamic mixing of synthetic air, 2 vol% H2 in air and certified mixtures of the contaminant gas in air. The response to 1 vol% H2 in air and is given by formula (1). This factor captures the increase or decrease of the sensor

response (as an apparent vol% H2) caused by the presence of

the contaminant. The XH factor is defined as the ratio of the Table 2 e Concentration of contaminants. difference in sensor response to 1 vol% hydrogen in air in the presence and absence of the interferent to the net sensor Gases Concentration [ppm] response to hydrogen in air and is given by the formula (2). Sulphur dioxide 500 The second factor reflects the change in sensitivity to Hydrogen sulphide 50/40a hydrogen in the presence of the interferent. Nitrogen dioxide 20 The third factor, X is defined as the ratio of net sensor HMDS 10 P response to 1 vol% H2 in air before and after the exposure to a Due to a change in gas bottle, the concentration of H2S was the contaminant and the net sensor response to 1 vol% H2 40 ppm for the test on the EC sensor. before the exposure. It is given by the formula (3). international journal of hydrogen energy 40 (2015) 11740e11747 11743

Fig. 1 e Exposure profile for the cross-sensitivity test showing hydrogen concentration (vol% in air) and contaminant concentration (arbitrary units). The letters a e c correspond to the different stages of the test as described in the text.

period. RH, final is the sensor response to 1 vol% hydrogen in

Ri R0 air in stage (c) as described in section Cross-Sensitivity X0 ¼ (1) RH R0 Testing Procedure.

R ; R X ¼ H i i 1 (2) H R R H 0 Results and discussion ¼ RH;final R0 XP 1 (3) X0,XH, and XP calculated for all the sensors tested are tabu- RH R0 lated in Table 3. According to the definitions, the contaminant R0 represents the sensor response in clean air; RH is the has no influence on the sensor response when the values for sensor response to 1 vol% hydrogen in air; Ri is the sensor X0 and XH are both equal to zero. The error of the cross- response in the presence of the interferent “i” (at concen- sensitivity factors was evaluated case by case considering trations listed in Table 2); RH,i is the sensor response to 1 vol the noise on the sensor response and the sensor resolution. In “ ” % hydrogen in air in the presence of the interferent i . case of XH, the error on the actual hydrogen concentration was These values are determined at the end of each exposure also taken into account. Both the sensor resolution and the

Table 3 e Cross-sensitivity factors and poisoning effect.

Species Sensor code X0 XH XP ± ± ± 50/40 ppm H2S Cat 1 0.06 0.01 0.05 0.02 0.05 0.01 Cat 2 0.02 ± 0.02 0.00 ± 0.02 0.03 ± 0.01 EC 0.04 ± 0.03 0.35 ± 0.05 0.04 ± 0.03 MOS 0.00 ± 0.03 0.01 ± 0.04 0.00 ± 0.03 ± ± ± 500 ppm SO2 Cat 1 0.84 0.02 0.08 0.04 0.45 0.02 Cat 2 0.00 ± 0.01 0.03 ± 0.02 0.03 ± 0.01 EC 0.07 ± 0.03 0.38 ± 0.05 0.11 ± 0.03 MOS 0.00 ± 0.03 0.01 ± 0.03 0.00 ± 0.03 ± ± ± 20 ppm NO2 Cat 1 0.00 0.01 0.01 0.02 0.00 0.01 Cat 2 eee EC 0.00 ± 0.03 0.01 ± 0.03 0.05 ± 0.03 MOS 0.00 ± 0.03 0.01 ± 0.03 0.00 ± 0.03 10 ppm HDMS Cat 1 0.00 ± 0.02 0.05 ± 0.03 0.04 ± 0.02 Cat 2 0.03 ± 0.01 0.03 ± 0.03 0.03 ± 0.01 EC 0.00 ± 0.03 0.01 ± 0.03 0.00 ± 0.03 MOS 0.00 ± 0.03 0.01 ± 0.03 0.00 ± 0.03 11744 international journal of hydrogen energy 40 (2015) 11740e11747

e Fig. 2 Effect of 50 ppm H2S on the response of the catalytic sensor CAT-1 (left) and the electrochemical sensor EC (right).

sensor lower detection level were determined experimentally. was much greater than the response observed for hydrogen

XP factors greater than 0.2 would be out-of-compliance with only (more than 30% increase of sensitivity). A pronounced the requirements of ISO 26142. drift of the signal can be observed. The MOSFET sensor response was not affected by the Sulphur-containing compounds exposure to H2S (see Table 3). Fig. 3 (left) shows the exposure profiles of the catalytic

sensors CAT-1 and CAT-2 to 500 ppm of SO2. A marked Sulphur-containing species such as sulphur dioxide (SO2) and decrease of the baseline was observed for CAT-1, which re- hydrogen sulphide (H2S) can be present as impurities either in hydrogen or in air, as sulphur is a common constituent in covers only partially following the removal of SO2.Asa many hydrocarbon fuels. consequence of the exposure to the contaminant, the sensor response to hydrogen decreases by 45% implying failure of the For the catalytic sensor CAT-1, the exposure to H2S resulted in a decrease of the baseline (see Fig. 2(left)). The effect was, sensor according to ISO26142. A minor increase of the net however, not permanent, as the sensor baseline returned to response to H2 was found during the exposure to SO2. Different behavior was observed for the other catalytic sensor the value prior to the H2S exposure after one exposure to H2 in clean air. The model CAT-2 was not affected by the exposure model. The sensitivity to H2 of CAT-2 was slightly decreased due to the exposure to SO2, but in this case the baseline was to H2S. For these sensors, at the given H2S concentrations, the unaffected. poisoning effect is negligible. The impact of H2S exposure on an electrochemical sensor is shown in Fig. 2 (right). The The effect of SO2 on the electrochemical sensor was similar to that of H2S (see Fig. 3 (right)). Both the baseline and the presence of H2S induced an increase in the baseline of the EC response to hydrogen increased. As observed for the presence sensor due to a positive cross-sensitivity to H2S. Remarkably, of H2S, the sensor signal is subject to drift. The effect on the the response of the sensor to a mixture of hydrogen and H2S

e Fig. 3 Effect of 500 ppm SO2 on the responses of the catalytic sensors CAT-1 and CAT-2 (left) and electrochemical sensor EC (right). international journal of hydrogen energy 40 (2015) 11740e11747 11745

± sensor response XH of 0.35 0.05 for 500 ppm SO2 is close the adsorption [26]. It was also found that the deposition of ± value of that 0.38 0.05 for 50 ppm H2S. This is a remarkable elemental sulphur on the Pt electrode was the main cause of result as the concentration of H2S is only 1/10th that of SO2. sensor degradation. This effect was reversible, i.e. the sensor

The response of the MOSFET platform to hydrogen was recovered to its normal sensitivity once H2Swasremoved. unchanged during exposure to SO2 (data not shown). These findings may help explain the pronounced, but tem-

Both H2S and SO2 can cause poisoning of the sensor cata- porary effect of H2S on the EC sensor. The gradual increase, lyst by blocking the active sites of the sensing element. The or drift of the signal could be indicative of a slow, secondary reactants, oxygen and hydrogen, are prevented from being electrochemical reaction taking place. The reversibility of adsorbed on the catalyst surface, which will influence the the effect is in line with reports regarding the effect of H2S response of the sensor. The higher poisoning activity of H2S on PEM fuel cells and on the H2S sensor. Further investiga- compared to SO2 has been explained by the electronic tion into the long term effects of the poisoning of EC sensors configuration of the SO2 molecule, which is attributed to the is necessary. shielding of the sulphur ion by oxygen [10,11]. The reaction of a sulphur containing compound with catalysts is the subject Nitrogen dioxide (NO2) of a significant research effort and has been described as a complex phenomenon involving both electronic and NO2 is a toxic gas produced in power plants, heavy industry, morphological changes [18]. and combustion processes including biomass burning. The

Among the sensor platforms studied, EC sensors as well as poisoning effect of NOx species on catalysts is subject to one model of the CAT sensors (CAT-1) were affected by debate and not yet well understood [27]. Numerous studies on sulphur-containing species. The design of the CAT sensor the effect of NOx on the performance of PEM fuel cell have models may be different with respect to the materials and been published (e.g. Ref. [22]), which employ Pt-based catalyst protective measures. While the exact nature of these are not materials. It has been proposed that NO2 is not poisoning the known (proprietary information), the observed difference in catalyst surface but rather affects the membrane or the cat- response to SO2 suggests more effective protective measures alystemembrane interface [28]. for the sensor model CAT-2. This engineering solution is also The sensor CAT-1 showed a decreased baseline and a reflected in the much higher price of the sensor. Another op- minor increase in sensitivity to H2 during exposure to NO2 tion could be the use of a different composition of catalyst, or (data not shown). The sensitivity of the sensor to hydrogen substrate material. The noble metal catalysts used on almost fully recovered once exposure to NO2 was terminated. hydrogen sensors have different susceptibilities to poisoning. The sensor CAT-2 was not exposed to NO2. The bonding of sulphur with metal withdraws electron The response of the electrochemical sensor was affected charge, which reduces the sticking probabilities and coverage by the presence of NO2, with the sensitivity to hydrogen of reactants on the catalyst [19]. The magnitude of this effect permanently reduced by about 5%. While NO2 can be consid- depends on the participation of the metal atom d-orbitals in ered as a poison for this platform (see Fig. 4), nevertheless the metal bonding (Pt < Pd), which indicates that Pd would be 5% change in sensitivity is within the allowable tolerances for more affected by sulphur poisoning than Pt. The effect of the poison resistance specified in the ISO 26142. catalyst support has been shown to be important, as some A slight effect of NO2 on the sensitivity to hydrogen of the materials may act as SO2 sinks [20]. A Pd catalyst deposited on MOSFET sensor was observed (data not shown), but the a carrier material capable of adsorbing SO2 tolerates higher magnitude of this effect is lower than the measurement un- concentrations of SO2 than on other types of material [21]. certainty (see Table 3) and therefore should not be viewed as To understand the poisoning mechanism of EC sensors, significant. studies on the effect of fuel impurities on the performance of PEM fuel cells can be useful (e.g. Ref. [22]). PEM fuel cell elec- trodes as well as the membrane are affected by poisoning species, this could also be the case for EC sensors.

The investigation of H2S chemisorption on Pt electrodes has revealed that H2S adsorbs as two types of sulphur spe- cies and, depending on the applied potentials, undergoes further electro-oxidation to elemental sulphur and sulphur dioxide [23,24]. For PEM fuel cells, transient concentrations of pure hydrogen may result in partial recovery of the cell performance [22] after exposure to sulphur-containing compounds. Although largely reversible, sulphur poisoning has been shown to be cumulative, as not all species of the adsorbates may be completely removed or could be further oxidized [25]. Chemical degradation of Nafion-type mem- branes can take place through formation of peroxide radi- cals, which form due to the presence of impurity cations. A report on an amperometric H2S sensor with a composite Pt e electrode also states that electro-oxidation products of H2S Fig. 4 Effect of 20 ppm NO2 on the response of the depend on the local electrode potential at the time of electrochemical sensor EC. 11746 international journal of hydrogen energy 40 (2015) 11740e11747

standard, including siloxane and the sulphur compounds. The

exposure of the CAT-1 sensor to 500 ppm of SO2 did, however, result in a marked lowering of the baseline, with potentially serious safety consequences. Moreover, this sensor model fails the resistance to poisoning test according to the ISO 26142 specifications. A recalibration of the sensor after expo-

sure to SO2 may be necessary. In all other cases the resistance to the selected species was good.

During exposure to H2S and SO2 the electrochemical sensor shows a remarkable increase of the sensitivity to hydrogen, i.e. by 35% and 38% respectively. This effect is notable even for

a 50 ppm concentration of the sulphur compound H2S. Sensor users should take this susceptibility of electrochemical sen- sors into account as it could lead to a false alarm. Also the drift of the sensor response to hydrogen during simultaneous

exposure to either H2SorSO2 warrants further investigation. Fig. 5 e Effect of 10 ppm HMDS on response of catalytic As the response of the sensor to hydrogen was not perma- sensor CAT-1. nently affected nor did the response even decrease during the exposure to sulphur compounds, these chemicals cannot be

properly classified as poisons for EC sensors. Exposure to NO2 Hexamethyldisiloxane (HMDS) at 20 ppm level caused a permanent decrease of the sensitivity

to hydrogen, in which case NO2 does act as a poison. Note, Silicon-based compounds are used as lubricants, sealants, however, that the sensor does not fail according to the and often as so-called inert ingredients in a variety of con- requirement of the ISO 26142 standard. sumer products. These compounds are recognized as poisons The MOSFET sensors seem to be the most robust sensor on several sensor platforms, including CAT. As mentioned type with respect to poisons. This may be due to an advanced above, the poisoning of catalytic sensors may be avoided sensor design, including use of filters and selective layers. through the use of filters, such as molecular sieves or acti- A more significant impact on the sensor response was ex- vated charcoal. Thicker filter layers prolong the lifetime of the pected for the tests with HMDS, as siloxanes are known to be sensor, but also lead to longer response times. As filter ma- harmful to sensors, especially the catalytic and electro- terials only have limited capacity to absorb contaminants, chemical platforms [30]. The impact observed on the catalytic other methods have been developed, such as a means to sensor, however, is well within the tolerance limits defined in desorb silicon adhering to the detection element. The the ISO 26142 standard, and the effect on the other platforms hydrogen sensor is exposed to a high concentration of was negligible with regards to the experimental error. The hydrogen, raising the temperature of the sensing element in siloxane resistance test prescribed in ISO 26142, based on order to remove any silicon compounds and recover full exposure to 10 ppm HMDS, may possibly not properly reflect sensitivity [29]. The effect of HMDS on the sensor performance sensor robustness to siloxanes. Further work on other Si-based has been shown to depend on the temperature, time and contaminants and other exposure profiles (e.g., concentration, concentration of HMDS. Poisoning of the sensor occurs exposure times) will be performed to evaluate the resistance of through the formation of a SiO2 layer the Pt/Pd surface [30]. sensors and to assess the relevance and suitability of HMDS as Although Si-based compounds such as HMDS are often re- representative species for Si-containing compounds. ported as catalyst poisons, none of the sensors investigated In general, these initial results of resistance to poisons show a significant degradation of performance upon exposure shown by the sensors are encouraging, as in almost all cases to HMDS at the concentration level and exposure time pre- the response to hydrogen is ultimately not affected. It should scribed by ISO 26142. HMDS is chosen as a suitable represen- be noted, however, that the exposures were of short duration, tative Si-containing species because of its availability, for single contaminants, and at a dry humidity (<5% RH). volatility and ease of use. Only in the case of the catalytic Increased humidity levels may result in a stronger effect of sensor CAT-1 was a minor change (increase) in the in the contaminants due to the formation of new species and sensitivity to H2 observed (see Fig. 5). moisture facilitated chemical reasons. Repeated or chronic During and after the exposure to HMDS, the response to low-level long-term exposures to poisons can be expected for hydrogen exhibited an increased drift. Longer exposures or sensor applications such as hydrogen refueling stations. An higher concentrations of HMDS are expected to be harmful to on-going study has revealed significant differences in the this sensor. Further investigations are needed to determine a performance of sensors deployed in the fields as compared to specific threshold concentration and exposure duration. those tested in the laboratory [31]. As the test parameters should be representative of the working environment of the sensors, further testing protocols should be developed to Conclusion ascertain the suitability of hydrogen sensors for more chal- lenging environments. The catalytic sensor platforms tested here were found to be Future work will focus on further assessment of the resis- robust towards most poisons identified in the ISO 26142 tance of sensors to contaminants by varying the testing international journal of hydrogen energy 40 (2015) 11740e11747 11747

protocols, e.g. the type of contaminant, concentration, expo- [14] Stokes, E.B. and J.Y. Gui, Gas sensor with protective gate, sure time and environmental conditions. The long-term effects method of forming the sensor, and method of sensing, 2001, of poisons may not be accurately reflected by comparing the Google Patents. [15] Firth, J.G., S.J. Gentry, and A. Jones, Catalytic gas detectors, response of the sensor before and immediately after exposure 1978, Google Patents. to the contaminant; therefore the definition of sensor dura- [16] Gentry SJ, Howarth SR. The use of charcoal and carbon cloth bility and resistance to poisons needs further discussion. for the poison-protection of catalytic gas sensors. Sensors Actuators 1984;5(4):265e73. [17] Wang, C.B. Tomasovic, B. Warburton, P.R. and A.Q. Wang, Poison resistant combustible gas sensors and method for Acknowledgements warning of poisoning, 2006, Google Patents. [18] Rodriguez JA, Hrbek J. Interaction of sulfur with well-defined This work was performed as part of an on-going collaboration metal and oxide surfaces: unraveling the mysteries behind between NREL and JRC and supported by the FCH JU project catalyst poisoning and desulfurization. Accounts Chem Res e H2Sense, GA number 325326 and Hyindoor for CAT sensors, 1999;32:719 28. [19] Feibelman PJ, Hamann DR. Electronic structure of a GA number 278534. The NREL Sensor Laboratory is supported “poisoned” transition-metal surface. Phys Rev Lett by DOE-EERE Fuel Cell Technologies Office. 1984;52(1):61e4. [20] Escandon LS, Ordonez~ S, Vega A, Dı´ez FV. Sulphur poisoning references of palladium catalysts used for methane combustion: effect of the support. J Hazard Mater 2008;153(1e2):742e50. [21] Lampert JK, Kazi MS, Farrauto RJ. Palladium catalyst performance for methane emissions abatement from lean [1] Boon-Brett L, Bousek J, Castello P, Salyk O, Harskamp F, burn natural gas vehicles. Appl Catal B Environ Aldea L, et al. Reliability of commercially available hydrogen 1997;14(3e4):211e23. sensors for detection of hydrogen at critical concentrations: [22] Zamel N, Li X. Effect of contaminants on polymer electrolyte Part I e testing facility and methodologies. Int J Hydrogen membrane fuel cells. Prog Energy Combust Sci Energy 2008;33(24):7648e57. 2011;37(3):292e329. [2] NREL Sensor Laboratory. [Accessed 17 April 2013]; Available [23] Lopes T, Paganin VA, Gonzalez ER. The effects of hydrogen from: http://www.nrel.gov/hydrogen/facilities_hsl.html. sulfide on the polymer electrolyte membrane fuel cell anode e [3] Boon-Brett L, Bousek J, Black G, Moretto P, Castello P, catalyst: H2S Pt/C interaction products. J Power Sources Hu¨ bert T, et al. Identifying performance gaps in hydrogen 2011;196(15):6256e63. safety sensor technology for automotive and stationary [24] Mathieu M-V, Primet M. Sulfurization and regeneration of applications. Int J Hydrogen Energy 2010;35(1):373e84. platinum. Appl Catal 1984;9(3):361e70. [4] Buttner WJ, Post MB, Burgess R, Rivkin C. An overview of [25] Sethuraman VA, Weidner JW. Analysis of sulfur poisoning on hydrogen safety sensors and requirements. Int J Hydrogen a PEM fuel cell electrode. Electrochimica Acta Energy 2011;36(3):2462e70. 2010;55(20):5683e94. [5] ISO 26142:2010 Hydrogen detection apparatus e Stationary [26] Yourong W, Heqing Y, E'Feng W. The electrochemical applications. oxidation and the quantitative determination of hydrogen [6] http://www.nrel.gov/docs/fy13osti/57207.pdf. sulfide on a solid polymer electrolyte-based system. J [7] Hu¨ bert T, Boon-Brett L, Black G, Banach U. Hydrogen sensors Electroanal Chem 2001;497(1e2):163e7. e a review. Sensors Actuators B Chem 2011;157(2):329e52. [27] Chen M, Du C, Zhang J, Wang P, Zhu T. Effect, mechanism [8] Korotcenkov G, Han SD, Stetter JR. Review of electrochemical and recovery of nitrogen oxides poisoning on oxygen hydrogen sensors. Chem Rev 2009;109(3):1402e33. reduction reaction at Pt/C catalysts. J Power Sources [9] Palmisano V, Boon-Brett L, Bonato C, Harskamp F, 2011;196(2):620e6. Buttner WJ, Post MB, et al. Evaluation of selectivity of [28] Mohtadi R, Lee WK, Van Zee JW. Assessing durability of commercial hydrogen sensors. Int J Hydrogen Energy cathodes exposed to common air impurities. J Power Sources 2014;39(35):20491e6. 2004;138(1e2):216e25. [10] Forzatti P, Lietti L. Catalyst deactivation. Catal Today [29] Tsukabayashi, S., H. Oishi, and K. Okajima, Hydrogen 1999;52(2e3):165e81. detection system, 2014, Google Patents. [11] Bartholomew CH. Mechanisms of catalyst deactivation. Appl [30] Matsumiya M, Shin W, Qiu F, Izu N, Matsubara I, Catal A General 2001;212(1e2):17e60. Murayama N. Poisoning of platinum thin film catalyst by [12] Korotcenkov G, Cho BK. Instability of metal oxide-based hexamethyldisiloxane (HMDS) for thermoelectric hydrogen conductometric gas sensors and approaches to stability gas sensor. Sensors Actuators B Chem 2003;96(3):516e22. improvement (short survey). Sensors Actuators B Chem [31] Buttner, W.J. Rivkin, C. Burgess, R. Hartmann, K. Weidner, E. 2011;156(2):527e38. Palmisano, V. et al., Impact of Environmental Parameters on [13] Chao Y, Yao S, Buttner WJ, Stetter JR. Amperometric sensor Hydrogen Safety Sensor Performance, in International for selective and stable hydrogen measurement. Sensors Conference on Hydrogen Safety 2015: Tokyo. Actuators B Chem 2005;106(2):784e90.